The cost of getting there slowly.

When owners tell me they're taking a wait-and-see approach to AI, I ask them one question: what exactly do you expect to see, and who do you think will show it to you? Because the market has arranged itself so that by the time the evidence is visible, the advantage has already been spent.

Here's the mechanic. A competitor who adopts early doesn't announce it. They don't cut prices either — not at first. They deliver the same work at a lower cost, pocket the margin, and spend it where you feel it last: better people, faster turnaround, sharper pitches for your clients. From the outside, nothing has changed. The quotes just got a little harder to win. The good hire chose them. By the time adoption becomes visible — when someone finally does cut prices — the early movers have had a year of compounding you can't buy back.

Waiting for proof means paying for it. The evidence you're waiting for is someone else's head start.

Why the learning can't be skipped

The deeper cost of slowness isn't the missed margin — it's the missed learning. Using these tools well is a skill, and skills compound. The team that started eighteen months ago isn't eighteen months ahead on a calendar; it's hundreds of experiments ahead on judgment. They know which work to trust the machine with, where it fails, how to check it, how to price what it changed. None of that transfers by reading about it. When the laggard finally buys the same tools, they buy the starting line, not the race.

And unlike previous technology waves, the tools don't hold still while you catch up. The gap between a firm learning at the release cadence and a firm deciding annually doesn't stay constant. It widens.

Fast is not the same as reckless

The honest objection is quality: we move slowly because our name is on the work. I take that seriously — so does every professional firm I advise. But look at where the risk actually sits. Moving fast on client-facing judgment: reckless. Moving fast on internal drafts, research, summaries, first versions that a human was always going to review: nearly free. The firms that get this right draw that line explicitly — aggressive inside the review boundary, conservative at the client boundary — and move the line as their judgment matures.

That's the discipline: speed where errors are cheap, care where errors are expensive, and no confusion between the two. Slow everywhere isn't prudence. It's just paying the highest possible price for the lowest possible learning.

The moat nobody can sell you

Most moats can be bought — locations, licenses, even talent. A two-year head start in organisational learning can't be. That's what makes speed of adoption the strangest competitive asset of this era: it's available to any firm of any size, it costs mostly attention, and every month it goes unclaimed it quietly transfers to someone else.

Getting there slowly still gets you there. It just gets you there as a customer of the firms that arrived first.